Developing a model of speech recognition process from an autopoietic approach

Abstract

Speech recognition is a time-tested research into the possibilities of reproducing human like qualities in machines such as computers. In the 40 years of exploration in this field, there are many milestones that have been set. Among the recent advances in speech recognition, the deployment of statistical techniques like Hidden Markov Models and connectionist techniques like Time Delayed Neural Network have rendered speech recognition into a very mature and commercially viable venture. Therefore, the direction of research in this paper is aimed at the improvement of current techniques rather than reinventing the wheel; particularly, autopoiesis is introduced in this paper to give a flair of self-organization to the cognitive process of speech recognition. It is hoped that this paper would provide complimentary alternatives to speech recognition rather than exclusive solutions it.